dplyr 基于过滤、分组和汇总计算新列

Question

鉴于下面的foo，我正在尝试根据y 值子集的平均值计算三组id 的异常y。在 dplyr 语法方面我看不到前进的方向，但必须有一个简单的方法。

也就是说，我想要一种方法来计算一个新列yAnom，即y-mean(y)，当x<=5为每个id时，y的值。基础 R 解决方案：

foo <- tibble(x = rep(1:10,3), y = rnorm(30), id = rep(letters[1:3],each=10))
# subtract y from mean of y for values of y when x <= 5 by group
#E.g., 
y5Means <- foo %>% group_by(id) %>% filter(x <= 5) %>% summarise(avg = mean(y)) %>% pull(avg)
foo$yAnom <- c(foo$y[foo$id == "a"] - y5Means[1],
               foo$y[foo$id == "b"] - y5Means[2],
               foo$y[foo$id == "c"] - y5Means[3])

帮助表示赞赏。

Answer 1

可能类似于以下内容：

foo %>%
  # new column with only the values you want to average over, and NA elsewhere
  mutate(y_when_x_under_5 = ifelse(x <= 5, y, NA)) %>%
  # new column with mean for each group
  group_by(id) %>%
  mutate(y_mean = mean(y_when_x_under_5, na.rm = TRUE)) %>%
  # required calculation
  mutate(yAnom = y - y_mean)